I'm working on a project where we want to try using pytorch-lightning to run rerankers on multiple GPU/TPUs. I can't seem to find any direct information if this is possible, as this link seems to only indicate that it can run on TPUs but without any information as to whether it is possible to run on multiple ones.
Hi! thanks for your contribution!, great first issue!
Lightning supports multiple GPUs/TPUs.
Just change gpus or tpu_cores to the amount of gpus or tpus you want to use
Hey @jhuang265,
tpu_cores: How many TPU cores to train on (1 or 8) / Single TPU to train on [1]
I am updating the doc to make it more clear :)
Best,
T.C